The Design And Implementation Of Recommendation System Based On Hadoop

When stepped into the21st century, we faced with era of Internet and the amount ofinformation grown rapidly. The big problem troubled us is information overload. In the past,our greatest engineers developed two way to deal with this problem, they are informationclassification and search engine. They did work and still be efficient. Yahoo and Sina did agreat job at information classification, while Google and Baidu helps people search any keyword. Sometimes people may be confused, and can’t think of a key word or category to lookfor the information. The recommendation system is considered as an excellent way to solvethis problem. Compare with former two ways, it is more intelligent.Recommendation system actively push the information to the users rather than enter keywords or think of a category. It based on collective intelligence and algorithm, and helppeople to find their required information. Collective intelligence is shared or groupintelligence that emerges from the collaboration, collective efforts, and competition of manyindividuals and appears in consensus decision making. Relies on huge amounts of data,recommendation engine analyze user behavior, characteristics and hobbies, and help users tofind the items that suit their interests.This paper first describes the research background, domestic and foreign researchsituation. And then studied the recommendation algorithm and its application. And alsoexplores the Hadoop framework and its principles. The paper determined the architecture ofrecommendation system, after that it detailed how to designed and implement the system.It has three contributions at this paper:1) Designed a recommendation system framework that can be expand at horizontal, andimplemented four engines.2) Using dynamic weight calculation method based on user feedback to combine theresults of the various recommendations from recommendation engine combination,. Itimproves the overall recommendation system evaluation3) Implement the recommendation system by big data platform Hadoop. As a result,the system increased computing efficiency and reduced the system’s response time.